| T O P I C R E V I E W |
| SiSL |
Posted - 15 March 2009 : 01:35:38
Okay guys, I'm stuck on a question now, please help me find out...
Now scenerio is:
1. There are people to vote for... Unnumbered...Lots Of...
2. Each people can be given from 1 to 5...
3. It is not necessary to vote for any of them...
4. You can vote for multi people at once..
So by end of the day, I should give those people a rating from 1 to 5 (or with decimals but not less than 1, not higher than 5) from checking their total votes, and number of people vote for each...
At end of the day, I'll have total number of votes given, number of votes for each voted person and sum of votes for each voted.
How do I do this rating?
|
| 6 L A T E S T R E P L I E S (Newest First) |
| SiSL |
Posted - 15 March 2009 : 14:24:22
Oh well, I found Bayesian Tech useful, I think following would ease my pain:
br = ( (avg_num_votes * avg_rating) + (this_num_votes * this_rating) ) / (avg_num_votes + this_num_votes)
avg_num_votes: The average number of votes of all items that have num_votes>0
avg_rating: The average rating of each item (again, of those that have num_votes>0)
this_num_votes: number of votes for this item
this_rating: the rating of this item
|
| Etymon |
Posted - 15 March 2009 : 13:53:11
SiSL,
I think you may need some stats nerds for this. I'm an ASP nerd who is familiar with stats, but I am not a stats nerd. You probably need some folks who do this sort of thing for some kind of self-punishing fun! LOL!
I did a web search for the keywords "statistics" "forum". The first result was this: http://talkstats.com ... that looks like a nice battle ground for your question.  The slogan in the footer of their site says, "Get free statistics help or just hang out and talk about statistics!"
In fact, I'm going to join that forum because I have some stats questions myself.
Cheers,
Etymon
|
| Etymon |
Posted - 15 March 2009 : 13:42:48
Ah, so you are worried about a weighted score where one person with less votes but high scores could get a better scores across the board in all categories. As in Person A has 5 votes with each vote having 4 stars in ever category and Person B has 300 votes with all kinds of stars in every category.
You are delving into statistics there my friend. I'll try to see what I can come up with. I have to fix a problem for Lon2 first, and then I can come over and visit with you on this one.
I think the word you are using "Attempt" means something like "situations where this person has helped someone" ... and the word "Voted" means something like "people who were helped who also voted".
|
| SiSL |
Posted - 15 March 2009 : 09:47:04
Here is example Etymon:
Person A: Got 4 Votes, Vote Total: 16, Avarage: 4,
Person B: Got 5 Votes, Vote Total: 20, Avarage: 4,
Person C: Got 3 Votes, Vote Total: 15, Avarage: 5,
Total Votes: 12 Votes, Total Points: 51
However, Even Person B and Person A got same rating, B Should be higher... For example, if one person got 1 vote for 4points, should not be higher than a person who got 5 votes avarage of 3.5 for example... Now imagine that with 300 people, with very different number of votes, number of total votes and try to give them a rating.
I am trying to do a Bayesian estimation, like how IMDB does it but no answer for which number to pick "required number of votes"
And sadly, when I find it, I had to work on it's SQL calculations :P
Okay real example
Person 1pt 2pt 3pt 4pt 5pt Attempt Voted
------------ -------- ------- ----- ------ ------ --------- ------
Person A: 40 21 48 80 204 2903 393
Person B: 5 5 8 9 36 787 63
Person C: 13 13 18 18 88 1552 135
(570 total)
--------------------------------------------------------------------
Total 806 318 582 868 3100 61212 6452
--------------------------------------------------------------------
Number of people voted: 17294 used 6452 votes.
How do you rank those Persons A, B, C from (1 to 5) relative to others... or would you seperately rank them?
Question to people who vote was: Rate this person according to help degree... |
| ruirib |
Posted - 15 March 2009 : 04:54:29
Why don't you just average the ratings obtained by each person, sum of Ratings for the person / number of ratings? I think that's what is usually done. |
| Etymon |
Posted - 15 March 2009 : 02:46:16
Can you give us some column names that you are working with in order to make up some queries for you?
|
|
|
